Multi-chassis node architecture is increasingly becoming a de facto deployment option for core optical network elements (e.g., optical routers or optical switches). This is mainly driven by the increase in need for high capacity network interfaces (e.g., Dense Wavelength Division Multiplexing (DWDM)) and the corresponding volume of routing and/or switching associated therewith.
Although a multi-chassis system includes multiple control processors, the Generalized Multiprotocol Label Switching (GMPLS) signaling protocol stack is typically designed in a centralized manner to avoid complexity and messaging overhead. However, under such a framework, a centralized GMPLS signaling architecture does not scale well because of the large number of Label Switched Path (LSP) tunnels and the bottlenecking that may occur.